Don't create invalid bitcasts.

[maleadt]

julia-debug: /home/tim/Julia/julia/deps/srccache/llvm-11.0.0/lib/IR/Instructions.cpp:2873: static llvm::CastInst* llvm::CastInst::Create(llvm::Instruction::CastOps, llvm::Value*, llvm::Type*, const llvm::Twine&, llvm::Instruction*): Assertion `castIsValid(op, S, Ty) && "Invalid cast!"' failed.

#4  0x00007ffff33e5d58 in llvm::CastInst::Create (op=llvm::Instruction::BitCast, S=0x5555582d6cf0, Ty=0x55555676b0a0, Name=..., InsertBefore=0x0) at /home/tim/Julia/julia/deps/srccache/llvm-11.0.0/lib/IR/Instructions.cpp:2873
#5  0x00007ffff73e0166 in llvm::IRBuilderBase::CreateCast (this=0x7fffffffa800, Op=llvm::Instruction::BitCast, V=0x5555582d6cf0, DestTy=0x55555676b0a0, Name=...)
    at /home/tim/Julia/julia/build/release/usr/include/llvm/IR/IRBuilder.h:2097
#6  0x00007ffff73e001b in llvm::IRBuilderBase::CreateBitCast (this=0x7fffffffa800, V=0x5555582d6cf0, DestTy=0x55555676b0a0, Name=...) at /home/tim/Julia/julia/build/release/usr/include/llvm/IR/IRBuilder.h:2056
#7  0x00007ffff736f255 in emit_inttoptr (ctx=..., v=0x5555583d1ef0, ty=0x55555676b0a0) at /home/tim/Julia/julia/src/codegen.cpp:1218
#8  0x00007ffff7394a5d in generic_bitcast (ctx=..., argv=0x7fffffff8b10) at /home/tim/Julia/julia/src/intrinsics.cpp:491
#9  0x00007ffff73978c8 in emit_intrinsic (ctx=..., f=JL_I::bitcast, args=0x7fff94c7a7b0, nargs=2) at /home/tim/Julia/julia/src/intrinsics.cpp:910
#10 0x00007ffff73a6102 in emit_call (ctx=..., ex=0x7fff94c63370, rt=0x7fff9b1570d0) at /home/tim/Julia/julia/src/codegen.cpp:3484

[vchuravy]

Can you add a lit test?

Also what happens when As != As && eType !== eType?

[yuyichao]

This is just an optimization so falling through is fine.
I assume this is for non-default codegen with normal non default address space?

[maleadt]

Default codegen, but with LLVMPtr:

(gdb) call jl_(lam)
map(typeof(CUDA.cudaconvert), Tuple{CUDA.CuArray{Float32, 2}, CUDA.CuArray{Float32, 2}, CUDA.CuArray{Float32, 2}}) from map(Any, Tuple{Any, Any, Any})

(gdb) call jl_(src)
Core.CodeInfo(code=Array{Any, (44,)}[
  Expr(:call, Base.getfield, Core.Argument(n=3), 1, true),
  Expr(:call, Base.getfield, SSAValue(1), :(:dims)),
  Expr(:call, Base.getfield, SSAValue(1), :(:baseptr)),
  Base.bitcast,
  Expr(:call, SSAValue(4), CUDA.CuPtr{Float32}, SSAValue(3)),
  Expr(:call, Base.getfield, SSAValue(1), :(:offset)),
  Base.add_ptr,
  Base.bitcast,
  Expr(:call, SSAValue(8), CUDA.UInt, SSAValue(5)),
  Expr(:call, Base.bitcast, UInt64, SSAValue(6)),
  Expr(:call, SSAValue(7), SSAValue(9), SSAValue(10)),
  Expr(:call, #<intrinsic #0 bitcast>, CUDA.CuPtr{Float32}, SSAValue(11)),
  Expr(:call, Base.bitcast, Core.LLVMPtr{Float32, 1}, SSAValue(12)),
  ...

I couldn't find an easy reproducer though.

[maleadt]

Both CI failures unrelated, and happen on current master too.